Free Demo
BETAFood Recognition API
Try ymove's food image recognition live. Upload a meal photo or describe what you ate: the AI identifies every food, estimates portions, and returns calories, protein, fat, and carbs.
0/2000
How the Food Image Recognition Works
Snap or type
Take a photo of your plate or describe your meal in plain language. No barcode scanning, no searching a food database by hand.
AI identifies every food
A vision model recognizes each item on the plate, estimates the portion in grams, and labels its own confidence. Unlike classic deep learning food recognition with a fixed class list, it handles mixed plates and dishes it has never been explicitly trained on.
Get calories and macros
Every item is matched against 2.9 million foods for accurate calories, protein, fat, and carbs, with totals for the whole meal.
Developers use the same endpoint to power an AI calorie counter, photo food logging, or macro tracking in their own apps. Everything you see in this demo comes back as structured JSON from one API call.
Food Recognition API FAQ
- Is there an API for food recognition?
- Yes. This demo runs on the ymove food recognition API. You POST a meal photo (base64) or a plain-text description and get back structured JSON: each identified food, its estimated portion in grams, a confidence label, and calories, protein, fat, and carbs per item plus meal totals.
- How does the food image recognition work?
- A vision AI model identifies every item on the plate, estimates portion sizes from visual cues, and matches each food against a nutrition database of 2.9 million foods covering USDA reference data and international packaged products. Because it uses a large vision model rather than a fixed-class deep learning classifier, it handles mixed plates, regional dishes, and foods it was never explicitly trained to label.
- How accurate is the food recognition?
- Each identified food gets a confidence label (high, medium, low) so your app can decide what to surface or double-check. Photo estimates depend on visible portion sizes and land close for common meals. Text input with quantities, like "200g grilled chicken with a cup of rice", is the most accurate because no portion guessing is needed.
- Can the API return calories and macros from a photo?
- Yes, that is the core use case: photo in, nutrition out. Upload a photo in this demo to see the exact JSON structure your app would receive, including per-item nutrition and meal totals.
- Does it also work with text descriptions?
- Yes. Users can type what they ate in plain language, like "two eggs, toast with butter and a glass of orange juice", and the API parses every item and calculates nutrition. Same response format as photo analysis, so your app handles both with one code path.
- How do I get started with the food recognition API?
- Sign up for an API key and call POST /v2/foods/log/photo or /v2/foods/log/text. The docs include request and response examples, and plans start with a free trial. This demo gives you 3 analyses per day without an account to evaluate quality first.
Add Food Recognition to Your App
The demo above runs on the ymove Nutrition API. Your users snap a photo or describe their meal, your app gets structured calories and macros back from one API call.